NDE & inspection

Automating nozzle weld inspection

23 April 2012

In July 2009, the European Commission awarded a two-year research contract under the Framework Program (FP7) theme to a consortium of eight partners from six European countries to develop an automated phased array inspection system to test complex-geometry nozzle welds in nuclear reactors. A prototype has been developed, and there are plans to commercialise the system. Patents are also being discussed. By Dimos Liaptsis, Vasilis Papadimitriou and Giannis Roditis.

The NozzleInspect project partners were: TWI Ltd (UK), KTU (Lithuania), CERETETH (Greece), Phoenix ISL (UK), PeakNDT Ltd, Vermon (France), Optel (Poland) and Iberdrola (Spain). Each of the partners brought their research expertise in different areas, such as ultrasonic simulation and phased array technique development (TWI and KTU), probe manufacturing, automated systems mechanical design and motion control, nozzle geometry and materials (Iberdrola), to help design the complete inspection system.

Complete inspection system
The complete inspection system on the full nozzle mockup during the demonstrations

Regular in-service inspection is important to verify the integrity of welded nozzle sections in nuclear and other safety-critical facilities due problems associated with thermal fatigue and stress corrosion. Early detection of cracks is therefore essential to ensure the continued safe operation of the nuclear facilities. However, the geometrical complexity of nozzle components and the changing surface curvature and weld profile around the nozzle circumference poses a number of challenges to the ultrasonic testing. These challenges are currently overcome by the use of multiple conventional ultrasonic probes that require frequent changes and calibration to cover the critical areas of the nozzle weld. Furthermore, complex mechanical manipulation is required that constantly adjusts the skew angle of the probe/s with respect to the weld in order to maintain the ultrasonic beam perpendicular to the inspection area. The probe changes are being carried out in a high radiation environment and that exposes the inspection personnel to harmful radiation and increases the overall inspection time.

In order to reduce the time and cost of such inspections there is a need to develop an inspection system capable of performing a full inspection of nozzle welds without the need to change ultrasonic probes. The main objectives of the automated robotic system design are:

  • Reduce the inspection times
  • Improve defect detectability and sizing
  • Reduce human intervention thus reduce radiation uptake
  • Reduce requirement for robotic manipulation that reduces size and cost of robotic deployment system.

Scanner design

To meet these objectives, the novel features of the developed scanner lie within two areas. First and foremost, its overall design offers a very high readiness level towards the inspection procedure, thus reducing by a great degree the preparation and calibration time—it is estimated to be 50% faster than the other method. Moreover, the inspection procedure is fully automated, under the parameters set by the operator from a remote, safe position away from the high radiation area of the reactor. Second, the flexibility that the scanner incorporates allows it to reliably adjust to nozzles of different diameters and to inspect nozzles that are not welded perpendicular to the cylindrical vessel.

To meet the first family of features, the assembly time in the inspection area was targeted to be drastically minimized, as was as any other need for zero-point calibration of the phased array (PA) ultrasonic probe. These guidelines concluded in the design of a fully integrated (one-piece) scanner that would simply need a quick, easy and safe latching system for mounting on the nozzle. All power and signal connections were also designed to be performed in a few fast steps.

The second family of features is served by two major developments in the design. First, an adjustable mechanism for the alteration of +/-20% of nominal diameter of the nozzle pipe (340 mm) is needed. A set of a coupled pair of linear actuators was selected for this task. Second, a flexible probe manipulator is needed to drive the phased array probe on the vessel surface, even in cases where the nozzle is welded on the vessel at non-perpendicular angles. A two-link serial manipulator was selected to accomplish this task. In addition to the above requirements, it is absolutely necessary to provide the PA probe with all the degrees of freedom that are required to perform the inspection in the optimum way. The degrees of freedom (d.o.f) are: a rotation of the probe at a full 360° peripherally around the nozzle, and a radial motion of the probe on the vessel, within a given range. The manipulator selected can reliably perform the aforementioned tasks in conjunction with a two d.o.f. gimbal joint on the probe end effector of the manipulator.

In practice, then, the system can be transported to the inspection area by two people and mounted on the feedwater pipe. Then, the operators need to latch the collar base, plug in the electrical connections and the air supply and ensure that all the safety pins of the system have been removed. All the system calibrations and the ultrasonic technique can be carried out remotely and away from the vicinity of the reactor.

Figure 1 shows a CAD model of the overall design of the automated system mounted on specific nozzle geometry that is commonly found in boiling water reactors (BWR). Regarding the motion-control aspects of the automated inspection system, the robotic scanner is able to accurately perform the inspection even if it is misplaced on the nozzle feedwater pipe, due to its advanced calibrating and controlling algorithms. There is a laser distance meter place on the first link of the manipulator and an inclinometer placed on the carriage. The scanner can be misplaced by several millimeters (tested at 50 mm misplacement) with respect to the vessel surface or peripherally by several degrees (tested at 20 degrees peripheral misplacement) with respect to the zero datum point on the vessel. The difference between perfect and actual placement of the robotic scanner is automatically measured by a number of sensors located on the scanner and the results are then fed into the control system. Using a hybrid position/force control algorithm, the control system takes into account the misplacement and accurately guides the robotic scanner to perform the inspection. For the control of the robotic scanner, a Galil DMC 3-axis controller was programmed to execute the low-level functionalities using the Galiltools software. The amplification as well as the controller interface of the systems sensors was made by a motion-control box designed and manufactured by PeakNDT.

Figure 1
Figure 1: At left, the nozzle and vessel; at right, the robotic scanner clamped on the nozzle

The robotic scanner's base consists of three main subsystems: the base frame, the clamping/centreing mechanism and the rotational degree of freedom. The base frame of the automated scanner is a lightweight yet rigid structure that supports all components and equipment. It is comprised of two 180° segments that are possible to unhinge to mount the robotic scanner on the nozzle feedwater pipe. The base is secured using a latch mechanism.

The clamping mechanism is located at the base frame and consists of four linear actuators, two manually- and two pneumatically-actuated (Figure 2). The purpose of the clamping mechanism is to stabilize the base on the nozzle and offer as well the required traction during operation. The two manually-adjustable actuators on either side of the feedwater pipe couple through a system of two timing belt stages and two low-backlash worm gears. As a result, the rotation of a hand knob control is transformed into an accurate coordinated linear motion of each linear actuator, adjusting the clamping diameter of the base on the specific nozzle pipe. The trapezoidal screw-nut configuration in conjunction with the very rigid steel linear guides provides high rigidity to the base. The adjustment of the manual actuators is made before the robotic scanner is moved on-site to avoid calibration near the vessel. After the robotic scanner is closed and secured using the latch, the base is concentric with the nozzle due to the geometrical restriction of coupled manual actuators. The pneumatic actuators are then activated remotely or from the scanner (via a control valve) and the robotic scanner is firmly clamped on the nozzle with a force of 30 N/m2 per clamp.

Figure 2: Latching and clamping operations of base
Figure 2: Latching and clamping operations of base. Timing belt that turns manual actuators shown at top

Robotic manipulator

The robotic manipulator is mounted on a carriage that can rotate 360° around the nozzle to fully cover the nozzle-to-vessel weld. Structurally, it is comprised of two links and is able to accurately place the end effector on the vessel in a workspace from 420 mm (on the weld centreline) up to 820 mm axially from the axis of the nozzle. The two links are selected to be of a rectangular aluminium profile in order to achieve high stiffness to torque and bending, while keeping the weight at low levels.

The power for both links is given by two 60 Watt brushed DC motors with planetary gearheads. The payload of the manipulator is estimated to 5 kg and it also able to apply a maximum of 40 N of force on the probe holder vertically to the surface of the vessel. This vertical force is essential to prevent the probe holder from flipping over as well as to ensure the compliance of the probe with the curved surface of the vessel. Two low-backlash 12:1-ratio worm gearboxes were also used due to low required inspection speed (maximum 50 mm/sec) as well as the requirement for accuracy and repeatability. Special care has also been taken in order to keep the inertia of the manipulator as low as possible. To achieve this goal, the power components of the manipulator were placed close to the carriage (Figure 3). To transmit power to the second link, a timing belt was used which also offered 1.5:1 reduction in rpm.

Figure 3: Two views of the two-degrees-of-freedom manipulator
Figure 3: Two views of the two-degrees-of-freedom manipulator

The end effector of the robotic scanner carries the probe holder of the 2D matrix phased array UT probe in a mount. During inspection, in order for the probe holder to comply with the varying surface of the vessel, two degrees of freedom are required. A gimbal joint was designed for that purpose that allows two axes of rotation perpendicular to each other. Rotation on one axis is powered by a timing belt; the probe swivels freely on the other axis. Figure 4 shows the design of the gimbal joint as well as the probe holder with the flexible membrane to conform to vessel surface curvature changes. The flexible probe holder was designed by Phoenix ISL.

Figure 4: Two views of the two degree-of-freedom end effector
Figure 4: Two views of the two degree-of-freedom end effector

During inspection it is required that a vertical force of 30N is applied to the probe holder to maintain constant ultrasonic coupling. Two springs mounted on linear shafts prevent the pressure of the manipulator against the vessel from exceeding tolerances, and to absorb variations due to vessel surface curvature. The deflection of the springs to known spring characteristics represents the vertical force at each time according to Hooke’s law. The deflection of the springs is measured using an absolute encoder (4096 ppr) and a two-link mechanism for converting linear to rotational movement. The precision achieved with this mechanism is 0.027 mm.

Phased array ultrasonic inspection

A number of requirements had to be achieved in development of the phased array inspection technique for the testing of the 139 mm-thick nozzle-to-vessel weld. The phased array probe must fulfill the following inspection parameters:

  • Centre operating frequency of 2 MHz, and operation in immersion
  • Electronically steer the ultrasonic beam between 35°–85°
  • Electronically skew the ultrasonic beam between 10°– -10°
  • Include a maximum of 128 elements
  • Have the ability to detect defects using direct skip because of the stainless cladding presence in the vessel inner surface, which significantly distorts the ultrasonic beam. Direct skip is used to send the ultrasonic beam directly to the weld without reflecting it at the inner surface of the vessel.

Several configurations were evaluated using a computer-based ultrasonic modelling package (CIVA 10) to assess the most suitable configuration, taking into account the inspection requirements and the manufacturability of the probe. Amongst these configurations are 2D matrix array and 2D annular segmented array. The modelling studies demonstrated that a 2D matrix array configuration at 2 MHz centre frequency with 128 rectangular elements (3 mm x 2 mm) in immersion is the most suitable configuration for this application. The resulting ultrasonic beam presents some grating lobes but these are low compared to the main beam and are not detrimental for the inspection under consideration.

The beam amplitude, focal area and focal depth were also modelled and confirmed as suitable for this 2D array configuration for the different operational configurations. It was demonstrated that the selected configuration gives the highest amplitude of the ultrasonic beam, the focal depths of -3dB and -6dB are the largest, and that the focal zone is able to reach the bottom of the inspected component. The capability of this array on electronic beam skewing was also assessed and a maximum skewing angle of +/-20° can be achieved. As a result, a less complex robot is required. The total improvement in defect detectability and sizing is estimated between 30-40%. The developed phased array probe was manufactured by Vermon.


For the laboratory phased array ultrasonic technique development, welded reference samples were manufactured with known sizes and position induced reflectors. The probe holder assembly was placed on the vessel of the nozzle reference samples as shown in Figure 5. Gel coupling was used between the flexible membrane and the sample in the laboratory development work, and water coupling was applied for the inspection of the real nozzle sample through the use of miniature spray nozzles. The profile of the nozzle-to-vessel weld varies circumferentially, so three reference samples were designed and manufactured to represent 0°, 45° and 90° parts of the nozzle weld profile and surface curvature profile.

Figure 5: UT nozzle test sample with weld profile at 0 degrees
Figure 5: UT nozzle test sample with weld profile at 0 degrees

Figure 5 shows the 2D matrix annular phased array probe placed on the sample representing the 0° part of the vessel. The sizes, orientation and position of the introduced notches into the reference samples were selected in accordance to ASME Code Case 235. The defects into the samples were introduced with Electro Discharge Machining (EDM). The skew angle of each defect ranges from -10° to 10° and the tilt angles are -5° and 5°. The defect size varies from 3.8 mm x 1.9 mm (L x H) for surface-breaking defects to 14.6 mm x 7.3 mm (L x H) for postulated defects. The phased array data was acquired and analyzed using the Micropulse 5PA (PeakNDT) phased array pulser-receiver.

Figure 6 shows the results of scanning the sample. The phased array data is displayed in corrected (left) and uncorrected sectorial scan (bottom right) and A-scan (top left). The A-scan displays the amplitude of the reflected ultrasonic wave of a specific incidence beam angle in the time range. The sectorial scan is a summation of individual A-scan data points and provides an overview of all the ultrasonic beam angles used to carry out the inspection. It shows that defect 3, positioned in the middle of the weld, dimensions of 5.8 mm x 2.9 mm, can be detected when the probe electronically skewed the beam at 5°. Also, the vessel inner-surface-breaking defect 5, with dimensions of 3.8 mm x 1.9 mm, is shown in the same phased array data display. Defects 3 and 5 are both tilted in the same direction. Suitable electronic skewing was applied to improve the detectability and increase the reflected amplitude response from the induced defects. There are some other strong reflections that can be seen in the phased array data. The first reflection that can be seen at 70 mm ultrasonic range is the reflection from the interface between the flexible membrane and the top surface of the steel sample. There are some other reflections in the sectorial scans that are due to internal ultrasonic reflections and reverberations occurring inside the probe holder immersion bath. These reflections did not affect the detection capabilities of the ultrasonic technique because they occur outside the region of interest.

Figure 6: Experimental results
Figure 6: Experimental results of the inspection of defect 3 and 5 in 0-degree reference sample

The inspection system prototype was tested on a full nozzle mockup with induced fatigue cracks in the nozzle-to-vessel weld (p33). The mockup used is an exact replica of an actual N5 feedwater nozzle that is used for qualification of scanning systems, ultrasonic techniques and inspection personnel. The system fitted in the constrained environment that exists in nuclear facilities. The robotic scanner was placed and clamped on the nozzle without any calibration or fine-tuning. The manipulator was able to move the ultrasonic phased array probe smoothly on the surface of the vessel and all the defects in the nozzle to vessel weld of the mockup were detected at the expected positions according to the defect plan and were accurately sized using the acquired phased array data.

CERETETH designed and built the scanning system and did all the programming for the motion control of the overall inspection system.

Author Info:

Dr Dimos Liaptsis (dimos.liaptsis@twi.co.uk) is an NDT Principal Project Leader within TWI NDT Validation Centre (Wales). Vasilis Papadimitriou and Giannis Roditis are Mechanical Engineers with Innora SA (www.innora.gr), but during this project they were with the Centre for Research and Technology Thessaly (CERETETH).

This article was published in the April 2012 issue of Nuclear Engineering International magazine.

Complete inspection system Complete inspection system
Figure 1 Figure 1
Figure 2: Latching and clamping operations of base Figure 2: Latching and clamping operations of base
Figure 3: Two views of the two-degrees-of-freedom manipulator Figure 3: Two views of the two-degrees-of-freedom manipulator
Figure 4: Two views of the two degree-of-freedom end effector Figure 4: Two views of the two degree-of-freedom end effector
Figure 5: UT nozzle test sample with weld profile at 0 degrees Figure 5: UT nozzle test sample with weld profile at 0 degrees
Figure 6: Experimental results Figure 6: Experimental results

Privacy Policy
We have updated our privacy policy. In the latest update it explains what cookies are and how we use them on our site. To learn more about cookies and their benefits, please view our privacy policy. Please be aware that parts of this site will not function correctly if you disable cookies. By continuing to use this site, you consent to our use of cookies in accordance with our privacy policy unless you have disabled them.